Taxpayers Protection Alliance Statement on Biden’s AI Executive Order

Earlier this week, the Biden Administration announced an executive order targeted at regulating Artificial Intelligence. In response, TPA’s Executive Director Patrick Hedger offered the following comments:

“The Biden Administration’s new sweeping executive order appears to be a stunning expansion of federal authority over Artificial Intelligence and offers more questions than answers,” said TPA Executive Director Patrick Hedger.“Preliminary concerns certainly exist in regards to overregulating AI to the point where American innovation and technology could be hindered, as well as issues surrounding the use of the Defense Production Act under the guise of ‘national security’ – as a crutch to defend pushing this administration’s AI policy priorities. Fundamentally, Congress should be legislating on this issue, rather than waiting for overreaching administrative actions from the current occupant of the Oval Office. Congress must take measured action in the AI sphere that implements a serious approach based in facts, not fear – America’s prosperity depends on it.”

Taxpayers Protection Alliance also developed a list of further questions from this week’s executive order announcement.

Questions About Biden’s Executive Order on Artificial Intelligence

President Joe Biden’s sweeping new executive order (EO) on artificial intelligence (AI) raises myriad questions. It likely marks a significant acceleration as Washington’s regulatory trash-compactor walls come together to crush industry and innovation. Blind to tradeoffs and opportunity costs, myopically fear-based AI policy making threatens to make the future American tech sector like the current European one: hyper-regulated, stunted, and largely irrelevant.

To what extent will the order dampen American innovation?

The White House’s fact sheet claims the EO “protects Americans‘ privacy, advances equity and civil rights, stands up for consumers and workers, promotes innovation and competition, advances American leadership around the world, and more.”

Voters should view skeptically any claim that a substantial regulatory proposal will produce vast good and no bad. The question almost never is Will Policy X do good or bad?, but rather Will the benefits of Policy X outweigh its drawbacks?As renowned economist Thomas Sowell famously says, “There are no solutions, only tradeoffs.” Although quite fearful of AI’s potential dangers, the Biden administration seems entirely bullish on the effects of government intervention in markets.

To take one example, calcifying industry safety tests (now administered by private actors) into formal regulatory reviews under the National Institute for Standards and Technology (NIST) will likely stultify the process of innovation and raise its costs. This would slow innovation and entrench the market shares of large incumbent firms (who possess the excess resources needed to navigate bureaucracy and compliance costs).

Why pander to special interests?

Biden’s EO contains numerous provisions that could insert left-wing policy biases into AI algorithms and the AI-fueled economy. According to the fact sheet, the EO will “support workers’ ability to bargain collectively,” a clear gesture to Biden loyalists in organized labor. To “[p]romote a fair, open, and competitive AI ecosystem” for “small businesses,” the White House invokes the Federal Trade Commission (FTC), which, under neo-Brandeisian, Biden-appointed Chair Lina Khan, has flouted decades of settled law as well as ethical norms to prosecute big businesses arbitrarily.

Moreover, the EO pledges to wield the Department of Justice to “address algorithmic discrimination,” a particularly concerning promise given Democrats’ propensity to judge discrimination based on “disparate impact” rather than “disparate intent” (as the Federal Communications Commission seems poised to do as well regarding the provision of broadband internet service).

How far should the government stretch national-security powers?

As noted by the R Street Institute’s Adam Thierer, the EO “stretches” the Defense Production Act to require that “companies developing any foundation model that poses a serious risk to national security, national economic security, or national public health and safety must notify the federal government when training the model, and must share the results of all red-team safety tests” (per the fact sheet).

Government officials often have abusively invoked terms such as “national security” and “public health” to effect ancillary policy priorities. Consider, for example, America’s national-security-predicated metal tariffs on the European Union and Canada, whose implementation and continuation reflect domestic political and economic factors, not geopolitical threats. Similarly, the Biden administration exploited and extended the Covid-19 emergency illegally to forgive student loans, suspend rent payments, and more.

The American public has little indication that the White House will interpret these terms any more narrowly or prudently in the context of AI than it has in others.

Can we dispense with the notion that no agency has existing authority to regulate AI?

AI doomers often justify sweeping new regulatory proposals by claiming that no agency has existing authorities to regulate AI. Biden’s EO should put to rest this unserious line of argument. While Congress has not designated an agency as a unified AI czar, myriad agencies obviously have extent powers that officials can apply to AI-related products and services (e.g., the FTC’s consumer-fraud powers). As FTC Chair Khan herself has said, repeatedly and quite correctly, “There is no AI exemption to the laws on the books[.]”

However, Biden’s order is quite sweeping, and the executive agencies that implement it must refrain carefully from exceeding their statutory bounds as they deal with AI. Should the Biden administration find its current powers insufficient to realize its policy goals, it may lobby Congress for more. The current Supreme Court has made clear that bureaucrats must operate within the confines of their statutory remit.